Volumetric changes in the amygdala and hippocampus are relevant to many disorders, but their close proximity makes it difficult to separate these structures by magnetic resonance imaging, leading many volumetric protocols to exclude problematic slices from analysis, or to analyze the amygdalo-hippocampal complex conjointly. The hippocampus tail is also often excluded, because of the difficulty in separating it from the thalamus. We have developed a reliable protocol for volumetric analysis and 3-D reconstruction of the amygdala and hippocampus (as a whole and in its anatomical parts). Twenty volunteers from clinical and healthy populations were recruited. T1-weighted images were acquired at 1.5 Tesla with native spatial resolution of 1.5 mm x 1.0 mm x 1.0 mm. Volumetric analyses were performed blind to diagnosis, using the interactive software package DISPLAY. Inter-rater (intrarater) intraclass correlations for the method were: 0.95 (0.88) for hippocampus tail, 0.83 (0.93) for hippocampus body, 0.95 (0.92) for hippocampus head, 0.96 (0.86) for total hippocampus and 0.86 (0.94) for amygdala. Volumes (mean+/-S.D.) corrected for intracranial volume for this mixed group were for the hippocampal tail: 0.325+/-0.087 cm(3); hippocampal body: 0.662+/-0.120 cm(3); hippocampal head: 1.23+/-0.174 cm(3); total hippocampus: 2.218+/-0.217 cm(3), and amygdala: 0.808+/-0.185 cm(3). In conclusion, the study demonstrates that the amygdala and hippocampal parts can be quantified reliably.